Bundling ZFS and Linux is impossible says Richard Stallman Free Software Foundation president and Gnu Public Licence (GNU GPL) author Richard Stallman has weighed in on the spat over whether Ubuntu can legally include ZFS in Linux, with a resounding “No!” Stallman has issued a statement he says “.... explains some issues about the meaning and enforcement of the GNU General Public …
"Illegal or impossible, not sure they are the same."
Either way, few people care now what Stallman thinks. Life is a lot more naunced that his rather black and white way of seeing the world. Linux will go in the direction that Linus and the users want it to go in, which almost certainly won't be the way this purist wants.
So how is this any different from, say, the non-free graphics drivers nVidia and AMD provide to power their graphics cards in Linux? Is it the fact it's designed as a baseline filesystem and therefore has to be low-level and integrated into the kernel (even as a module--kernel modules have to match the kernel so are part of the tree) to do that that's the issue?
It's a piece of cake: no Linux distro is allowed to distribute Nvidia binary drivers. End-user is the one who actually search for, downloads and installs the driver and this is compliance bothe with GPL and Nvidia proprietary licenses. It it inconvenient but GPL crowd also respects other licenses too.
Remember, we're talking about the permission to distribute not the permission to use.
It's a piece of cake: no Linux distro is allowed to distribute Nvidia binary drivers. End-user is the one who actually search for, downloads and installs the driver and this is compliance bothe with GPL and Nvidia proprietary licenses. It it inconvenient but GPL crowd also respects other licenses too.
Remember, we're talking about the permission to distribute not the permission to use.
That doesn't sound quite right. No licence can prevent one exercising one's right to free speech. Otherwise i couldn't mix GPL code with, on the same CD, the works of Shakespeare, a list of English words or indeed the file allocation tables off the media itself.. That would clearly be nuts.
Clearly you can distribute GPL licensed software alongside non-GPL files, no matter what the GPL says.
"That doesn't sound quite right. No licence can prevent one exercising one's right to free speech. Otherwise i couldn't mix GPL code with, on the same CD, the works of Shakespeare, a list of English words or indeed the file allocation tables off the media itself.. That would clearly be nuts."
Um, the license is related to copyright, and copyright IS a legally-erected restriction on the First Amendment: namely, speech and press can have ownership, and that ownership can impose restrictions which are supported by the law.
The works of Shakespeare are generally OUT of copyright because Shakespeare died hundreds of years ago. English words fall under the "too generic" escape clause (but a compiled dictionary can be subject to copyright), and the file allocation tables are the product of machine, not man.
"Clearly you can distribute GPL licensed software alongside non-GPL files, no matter what the GPL says."
No, because you can violate the license and copyright associated with the GPL. Without copyright, authors can't prevent their works being copied and so on.
"no Linux distro is allowed to distribute Nvidia binary drivers"
Oh really?
pfred1@buck:~$ apt-cache search nvidia
nvidia-glx - NVIDIA metapackage
nvidia-kernel-source - NVIDIA binary kernel module source
nvidia-kernel-686-pae - NVIDIA kernel module for Linux (686-pae flavor)
nvidia-driver-bin - NVIDIA driver support binaries
The GPL license only affects code licensed under the GPL. The GPL CANNOT restrict you from distributing software with other licenses.
First, the graphics drivers were specifically written to interface with Windows.
Second, the interface between the GPL kernel and the proprietary graphics drivers is also GPL.
Third, it is the user that obtains the driver directly from the proprietary vendor, thus gains permission from that vendor.
Sorry but you are plain wrong. That black and white enough for you.
This isn't a matter of what Stallman thinks but what the license says. Whilst you may disagree with Stallman (and many people do, though many people think he's done rather a lot of good for software), his reading of the license is probably correct. Subtle nuances are really nice in music and art, in license terms they are a sodding nightmare.
I suspect Ubuntu found legal advice that says it *may* not be illegal due to this tenuous legal thread that we've worked up and devised. I'd be astonished if a lawyer went out on a limb and said it was legal. Lawyers never commit to anything like that even if they are 100% certain they are right.
I'm thinking back and can't remember a case of somebody winning against the GPL2 license. I could be wrong though.
When Stallman speaks, my advice would be to listen carefully, You may not like what you hear, but as one of the main architects of free/open software AND the GPL license he tends to carry a lot of weight (no pun intended).
"Sorry but you are plain wrong. That black and white enough for you."
Hi there, A/C.
Are you a judge who's had this case to decide? Because if you're not neither your view nor Stallman's nor Ubuntu's nor Linus's nor my opinion counts.
The contention on Ubuntu's side seems to be that loading a module does not create a derived work of the kernel, it's just two works sitting alongside each other in memory. Until we get a court ruling on that we don't know whether it's right or wrong. Stallman may have been responsible for the GPL but even he can't determine how the court will rule on the facts of a specific case.
This post has been deleted by its author
"The problem is not loading a module, it's the legal permission to distribute that module with Ubuntu or any other distro."
The solution being NOT to distribute it. As with the other non-GPL stuff the distro package is a loader that pulls the real package down from a separate resource.
Problem solved.
"Are you a judge who's had this case to decide? Because if you're not neither your view nor Stallman's nor Ubuntu's nor Linus's nor my opinion counts."
You are correct, of course, and that should make us all fearful.
If my I learned anything from my old Intellectual Property professor in law school it was that the courts simply don't understand IP law. Prof. Blaustein proved that by forcing us to read every off-the-wall, counter-intuitive and contradictory decisions made by the US Supreme Court up to passage of the 1976 Copyright law (I took the class in 1980). And most of those involved technology that most today would consider backward, like gramophone records. Operating systems? The Internet? What? I (proudly) don't even know how to send an e-mail!
We are all so screwed.
Yes, it does.
Kernel modules do not exist in a bubble, they're directly linked (in the compiler sense) to Linux, which is licensed under the GPL. Those modules are therefore inescapably "derivative works" of that GPL licensed code, as they literally contain parts of it, simply cannot function without it, and therefore must be published under the same license, or else be in violation of copyright law (assuming they are actually published, as opposed to used privately).
The question is not whether or not this is a copyright violation, as it quite clearly is, the question is whether or not the copyright holders (and in Linux there are many) have any interest in challenging this particular violation, and in this case I suspect they won't.
As the author of the GPL, I'm fairly confident that Stallman probably understands it well enough to know when it's being violated. The fact that certain Linux (and other Free Software) developers seem not to care, is not somehow a poor reflection of Stallman, but rather a stark example of the sort of hypocritical "pragmatists" who are quite happy to benefit from the environment of academic freedom created by this license, but then whine about "fundamentalism" (as one of the more diplomatic examples of their rhetoric) whenever they feel the urge to prostitute themselves to some proprietary software vendor, whilst sacrificing our freedom at the alter of their "convenience".
"When Stallman speaks, my advice would be to listen carefully,"
I've listened to him enough over the years and frankly I can't be bothered anymore. Some would call him an OSS "evangelist", I'd call him an inflexible arrogant zealot. If you want to listen to his quasi religious diatribes that are usually one step removed from reality then knock yourself out. For me - I'll just get on with using software that suits my needs, regardless of its provenance or license conditions.
"I've listened to him enough over the years and frankly I can't be bothered anymore"
I've listened to people like you enough and cant take it anymore. How the hell did you come to the conclusion that RMS has ANYTHING to do with Open Source Software? He has never been involved in OSS has wont touch it with a barge pole. It's like complaining that Santa failed to deliver the easter eggs on easter sunday. Or complaining to the grower of apple trees that the tree he sold you did not produce any bananas.
This is a political issue. If you dont like politics fine, continue to avoid voting.
@JeffyPooh:
Not even "illegal".
Violating a contract can open up civil liability, but it's not 'illegal'.
That point is interesting. Even if rms got his way in court, what damages could he claim? It's not like he could claim lost income, the entire point of GPL is that it's free. Whatever gets bundled with GPL code, it can't cause him material loss, so it basically boils down to hurt feelings. How do US courts deal with that? It's not like the US has european style "moral rights".
It might not cause HIM material loss, but it could very well harm the GPL2 eco-system and cause others harm. It wouldn't be the first time that someone sought their fortune by claiming ownership of Linux because their code was included in it by a third party. *cough*SCO*cough*
We had these exact arguments years ago. In the US, a creator has copyright and may restrict use and derivative uses of her creation. The law governing infringement is civil law, but there are instances when, if one infringes upon the wrong thing, the FBI comes and arrest you.
Commercial software writers specify the terms of use in an EULA as per their rights as a copyright holder, and disputes over an allegedly unlicensed use are pursued against purchasers under contract law.
No one has permission to use GPL software, or any software under copyright, except they receive a license. There is an allowance for fair use, and while ZFS on Ubuntu may be rationalized under that basis, I expect that a Court would have to decide. I expect that Canonical's positive advice hinges on ZFS being licensable by the end user and that Canonical does know how to distribute non-GPL software to users.
My personal opinion? I can license GNU software to run on my FreeBSD and OS X systems. If a Linux distributor wants to include a non-GPL file system to improve integrity and function for the user, it should be possible without one author coercing another author to change his license. As the catalyst for the GPL was a vendor's withdrawal of code that Stallman was using at MIT, it would be ironic to use the GPL to have Canonical withdraw its packaging of ZFS. I think Stallman should let this be, because if the slippery slope is to more user choices and inclusion of better technology, let's slide. GPL software's importance will not be diminished.
That point is interesting. Even if rms got his way in court, what damages could he claim?
Damages as in monetary damage? Probably none, at most very, very little.
Given that the properties involved here would be the Linux kernel, property of the Linux Foundation, and ZFS, property of Oracle, I'd guess he doesn't even have standing to bring a suit in the matter at all. I suppose you could make some tenuous thread between the GNU utilities being distributed along-side ZFS, but the actual, tangible link is only between the kernel and the file system, neither of which are FSF properties. I think any suit Stallman tried to bring here would get thrown out entirely.
But let's pretend for a moment that he did have standing to bring a suit or that it was Linus (who, as the public face of the Linux Foundation, probably would have standing) calling foul on Ubuntu. At that point the case would become a lot clearer and the possible end result would be an injunction, not damages.
"Damages" are only one possible part of a legal remedy.
As you indicate, the "damages" in this case would probably be limited to legal fees, since there is no income to be deprived of.
However, the remedy would probably also include an injunction against the prohibited behavior, i.e. either they bundle with all source code, or they do not bundle at all. That is the remedy RMS would be most interested in. You can not bundle non-free software with free software with impunity; that part of the license has been litigated before.
Aspects of 'hurt feelings' or 'moral rights' do not have obvious relevance in this scenario.
"the entire point of GPL is that it's free"
No its not. It has nothing to do with being free. The GPL has everything to do with being free!
Confused? Shame about our language. I'm assuming you are using free = zero cost. Well the GPL is not zero cost. Plenty of GPL software is happily sold for a nice profit my many many companies every day. I once bought RedHat, off a store shelf.
People buy my GPL'ed software all the time. But they are free because of it. ZFS bundled into Ubuntu removes that, or at least threatens that. The GPL is designed to protect that freedom.
You need to be careful of your definitions here. The GPL may usually be free, but does not need to be free as it places no restriction on charging a price but the GPL is entirely free, all of the time and guarantees preserving that freedom to anyone who has the GPL'ed code.
Just invest further in refining BTRFS. It still hasn't reached the maturity of ZFS, but it's stable enough for production use now, and storage management is a whole lot easier than ZFS - you can pull a drive from a volume with ease, unlike ZFS. It just needs a few more features, mostly relating to performance.
If you are running over a cluster then IBM's GPFS can knock ZFS into a cocked hat. You have to endure IBM's sale process though and pay them silly money.
Re: BTRFS
One of the key benefits of BTRFS over ZFS would be over the design and API. Whilst everyone has been focusing on the ZFS source, we should not forget the other battles involving Oracle and specifically those around API ownership and the IPR embedded in the design of ZFS - remember MS and long filenames...
However, depending upon the IPR issues around design and API's it might be relatively easy to use these and using a clean room approach, create new code implementing ZFS...
No, dont split effort doing the same thing for multiple implementations.
Yeah, screw competition and innovation; let's just cede all technological progress to multi-billion-dollar international mega-corporations, with histories of ridiculous litigation and screwing over communities, in pursuit of ever-larger yachts.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
